Research On 3D Reconstruction Algorithm Based On Aerial Image Of Unmanned Aerial Vehicle

Under the background of expansion of the digital equipment,how to make people feel the three-dimensional sense of the objective world in the virtual world has become a research hotspot.At the same time,due to the rapid development of drones in recent years and their flexible and convenient features,the application fields also follow more and more,so the three-dimensional reconstruction of the aerial image of the drone has important application value in the outdoor scene.The traditional three-dimensional reconstruction based on aerial image less considers the impact of weather on image quality.However,in recent years,due to air quality declining,haze weather has caused great interference to human life,and the image obtained by aerial photography is also inevitably affected.Therefore,it is very important to defog the aerial image;at the same time,feature point extraction and matching is the basis of three-dimensional reconstruction and the key to the success of three-dimensional reconstruction.Point extraction and matching is the basis of 3D reconstruction and the key to the success of 3D reconstruction.Because of the existence of void phenomena in 3D reconstructed point clouds,it is also important to study cavity repair in order to make the reconstructed point cloud effect more intensive.The main work and innovations of this thesis are as follows:(1)Research on integrated defogging optimization techniques based on dark channels.Firstly,this thesis studies the basic theory of image fog removal,then according to the fog imaging model chooses dark channel defogging algorithm,finally,based on the dark channel prior theory,proposes a comprehensive optimization algorithm for aerial image fog removal,which respectively to dark channel simulation programming,atmosphere and light transmittance is optimized,and the integrated optimization of the processing.In order to verify the superiority,feasibility and universality of the proposed algorithm,several comparative analysis experiments are carried out.(2)Research on the feature extraction and matching scheme based on PCA processed image.By extracting the main components of the image,the method shortens the experimental time while highlighting the desired scene.This method highlights thescenes that need to be reconstructed and improves the reconstruction effect.In this thesis the basic principles of classical feature extraction and matching algorithm and PCA algorithm are introduced respectively.Secondly,FAST algorithm,Harris algorithm and SIFT algorithm are used to extract and match the feature points of the defogged image,and the comparison experiment is carried out to verify the applicability of SIFT algorithm.Then the contrast experiments of feature extraction and matching on the images before and after the PCA algorithm are added to prove the effectiveness of the algorithm.Finally,to verify the feasibility of the proposed algorithm,several groups of different images are experimented and analyzed..(3)Research point cloud cavity repair algorithms,including hole boundary extraction and hole filling repair.Firstly,according to the basic theory of 3D reconstruction algorithm,the SFM algorithm and PMVS algorithm are mainly described,and the 3D reconstruction process is designed and analyzed.Then the cavity repair theory is analyzed for the cavity phenomenon of the reconstructed point cloud.The experimental results verify the feasibility of the proposed algorithm.The applicability and practical application value of the proposed algorithm are verified.